What 5.5M interviews reveal about scorecard completion

Only 31% of candidate interviews ever get a scorecard at all, and of the scorecards that do get created, just 32% have every field filled while 26% come back completely empty.¹ That gap explains more about noisy quality-of-hire data than any dashboard ever will. The evidence your hiring decisions depend on is sitting in interviewers' heads, decaying by the hour, and most of it never makes it onto paper.

Here's the counter-number. When Metaview's AI generates the scorecard from the interview audio first, the economics flip. Scorecard completion jumps by roughly 60%, from 31% to 50.3%, and the drafts arrive with roughly 3x more fields filled, 7.85 against 2.61.¹ Same interviewers, same roles, same rubrics. The only thing that changed is who writes the first draft, and recruiters stay in charge: Metaview captures the evidence, and interviewers add the judgment and sign off on every field.

This post breaks down where the completion gap comes from, what 5.5 million captured conversations show about closing it, and what genuinely complete scorecards let a talent acquisition leader do that half-empty ones never will.

Why scorecards go unfinished

Scorecards compete with everything else in a recruiter's day, and they usually lose. The workflow asks for 30 to 45 minutes of unpaid admin after a meeting that already consumed an hour, and the next calendar block starts in five. By the time there's a free moment, the details have started to fade, and the scorecard rubric is still sitting empty.

The cost shows up later, where it's hardest to argue with. Debriefs run on vibes because half the panel submitted two lines and a rating. Quality-of-hire analysis turns circular: you can't connect interview signal to post-hire performance when the signal was never written down. And the scorecards that arrive a week late read as reconstructions, drafted from a memory that's already merged three candidates into one.

The instinct is to fix this with policy. Completion mandates, reminder sequences, dashboards that name and shame. Teams that run that play get a bump for a quarter, and then the numbers slide back, because the underlying time cost never moved.

What changes when the AI drafts first

The corpus data says the fix is structural. When the AI generates the draft first, the scorecard gets submitted 50.3% of the time instead of 28.6%, and it comes back with 7.85 fields filled instead of 2.61.¹ No change-management program needed: the time cost of completing a scorecard drops from half an hour to a few minutes of review.

Read the asymmetry carefully, because it's the whole story. The interviewers didn't get more diligent, and the rubric didn't get shorter. The draft stopped being their job, so the part that is their job, the judgment, finally got done. The gap between 2.61 filled fields and 7.85 is the distance between asking people to be stenographers and asking them to be evaluators.

The mechanism: capture first, structure second

Metaview's AI Notetaker captures every spoken word of the interview, which means the scorecard drafts itself against the rubric, the debrief surfaces direct quotes instead of paraphrases, and the hiring manager reads evidence within the hour instead of a summary three days later. The interviewer opens a pre-structured draft, corrects what the AI got wrong, adds the judgment only a human can add, and submits. AI drafts and organizes. Humans evaluate and decide.

Template fit matters more than teams expect. A phone screen, a technical deep dive, and a panel debrief produce different kinds of evidence, so the meeting type is auto-detected from the calendar and the notes structure follows it. Choose the template that matches the conversation, or build a custom one, and the draft lands organized by your questions and answers rather than as a wall of transcript.

The downstream numbers move together. According to Metaview's 2026 AI & Hiring Alignment Report, surveying 505 recruiting leaders and hiring managers across North America and EMEA, teams that put AI at the core of hiring don't just document more. They screen more, decide faster, and hit goals more often.

What TA leaders do with complete scorecards

Complete scorecards matter because of what they make possible: everything you've wanted to do with interview data and couldn't, because the data didn't exist. When the evidence is captured and structured instead of half-empty, the options open up:

• Calibrate interviewers on evidence. When every rating ties to captured quotes, you can see which interviewers consistently surface signal and which run charming conversations that produce nothing, the gap covered in good interviewer, bad interviewer.
• Run debriefs on the record. Disagreements resolve by checking what the candidate said instead of whose memory wins. Decisions speed up because the argument surface shrinks.
• Connect interviews to outcomes. Quality-of-hire models stop being aspirational when the input side of the model is captured and structured instead of two-thirds empty.
• Defend decisions later. A complete, timestamped evidence trail beats a reconstructed one in any audit, debrief, or candidate dispute.

The teams furthest down this road treat scorecard data as an operating asset rather than HR paperwork.

That's what capture-first scorecards buy at the team level: the documentation tax converts back into recruiter capacity, and the evidence base compounds with every interview.

Measure your own completion rate this quarter

You can't manage a number you've never measured, and most teams have never measured this one. Pull your last 90 days of interviews from the ATS and count how many have a fully completed scorecard. If most of them don't, you've found the cheapest quality-of-hire improvement available to you this year.

Then instrument it properly. Metaview Reports tracks completion and feedback latency across every interviewer and role, and feedback speed is one of the levers that shows up in time-to-fill within a single quarter. The scorecard is the data layer your next hundred hiring decisions run on.

¹ Source: Metaview's corpus of 5.5 million captured conversations (5.2 million candidate interviews), 2026. Scorecard completeness, field counts, and AI-generated versus manual submission rates drawn from submitted-scorecard samples on the platform.